Edit model card

Returns common English accent given a voice audio sample.

See https://www.kaggle.com/code/dima806/common-voice-accent-classification for more details.

image/png

Classification report:

              precision    recall  f1-score   support

          us     0.3956    0.0150    0.0290      4788
     england     0.5255    0.9121    0.6668     18082
      indian     0.5883    0.4586    0.5154      5656
   australia     0.4962    0.0381    0.0707      5124
      canada     0.3714    0.1760    0.2389      5169

    accuracy                         0.5220     38819
   macro avg     0.4754    0.3200    0.3042     38819
weighted avg     0.4942    0.5220    0.4304     38819
Downloads last month
73
Safetensors
Model size
94.6M params
Tensor type
F32
·